Wind turbines have historically suffered various recurring failures. Some have been eradicated, but other breakdowns continue to plague the industry. Gearbox reliability is one such issue. Early failures were mainly attributed to micro-pitting in gears (or grey staining) but that is now almost eliminated. Bearing failures are now the most prevalent issue. Various vibration monitoring systems, for detecting bearing problems have been tested, but not with complete success. Detection problems come from slow-shaft speeds, planetary sections, and varying rpm and loads. Owners have found vibration analysis good for detecting cracked or missing teeth, mis-alignment, or unusual resonance. However, the technique often only shows bearing failure at an advanced stage. While useful in a predictive sense, it doesn’t proactively prevent failure.
Oil analysis has been proven in research and use that it can detect tell-tale wear particles produced at early stages of gear and bearing failure. Researchers at Monash University in Australia note that the first signs of machine failure can be in 1/15th of the time needed by vibration analysis to reveal the same conditions.
Until now, however, oil analysis has had limited success in the wind-energy industry. With limited access to wind turbines, oil samples are drawn infrequently. Easily accessible industrial gearboxes should be sampled anywhere between 300 and 1,000 hours, depending on reliability goals. Wind-turbine gearboxes, on the other hand, are usually sampled at much longer intervals, often up to 4,000 hrs or more. Consequently, the benefit of early warning is lost and the problem progresses to complete failure.
On-line analysis sensors are one solution to the problem for several reasons. For instance, they:
• Maximizes information for best maintenance planning, allowing a longer lead time on planning and eliminating unnecessary preventative activity during the time up the tower.
• Provides real-time application data, so OEMs get better insight to their equipment, and operators can instantly access information on the machine or oil health.
• Provides a low-cost strategy to equip all critical systems, and ensures consistency of data across all units.
• Reduces current manpower demands and minimizes operator exposure to safety and health hazards.
• Provides an ideal screening tool to eliminate need for unnecessary oil sampling thus allowing for laboratory intervention on exception alarms only.
• Provides more than just a protection system to shut down the unit in the event of possible catastrophic failure.
• Minimizes transport and disposal of oil samples, and eliminates use of reagents in oil-analysis laboratories.
Sensors for oil monitoring fall into several areas, such as:
Dielectric sensors for oil condition and moisture contamination
These monitor oil quality to ensure it is fit for purpose. The sensors for doing so are often based on measuring an oil’s dielectric properties. Such sensors have been used successfully in other industries. This type of sensor will detect oil oxidation as well as gross water contamination. A variant is the moisture sensor which uses a water-soluble membrane to measure the relative humidity of the oil or airspace. Water is one of the most critical contaminants in a wind-turbine gearbox. Small amounts can significantly shorten gear, bearing, and oil life. Monitoring relative humidity ensures detecting increasing levels of moisture contamination long before it becomes more damaging.
Laser and optical sensors in particle counters
Particle counting is a necessary root-cause measurement. These sensors are similar to optical particle counters found in laboratories. Such a unit measures a small portion of the oil and reports the contamination per ISO 4406:1999 (Quantities of particles/ml of fluid in the >4µm, >6µm and >14µm ranges).
This information can lead to the early identifi-cation of gear and bearing failure. Limitations associated with on-line optical counters include the cleanliness, aeration, water contamination, and oil color. Conventional wind turbine gearboxes have little sump volume, and use oils sufficiently viscous to keep trapped air from escaping in a short period.
Unfortunately, air bubbles and water droplets can be erroneously counted as particles. Early trials showed these sensors not very successful in the field for this reason. However, manufacturers have been developing air bubble and water droplet detection, and even de-gassing versions.
Mesh and pore sensors monitor solid particles and other debris
Unlike optical sensors, pore-blockage monitoring devices are not affected by water, aeration, color, or even extreme levels of solids in oil. The downside is that, by ISO definition, the units are not considered particle counters. In addition, the units are not solid state in operation, making them more complex. As a result, they are more expensive and require periodic servicing. However, these are a more reliable alternative to detecting particles as small as 4µm.
Magnetic detection for full flow wear debris sensors
Using magnetic detection principles, metallic wear debris particles can be detected thus indicating bearing or gear problems. Such systems have been used in military and commercial aviation for many years. These sensors can be placed directly in the oil flow before the filter. This allows the monitoring of all the circulating oil, not just a portion of it. The problem with this set up is poor sensitivity. The largest bore sensors detect only particles larger than 350µm (0.35mm). This size of wear debris is considered large and normally indicative of imminent failure. Thus these are of little value in detecting minor changes in wear rates.
Magnetic detection for partial flow wear-debris sensing
Using similar technology to the full flow version, partial-flow variants measure smaller particles (>40µm) normally associated with the initial stages of changing wear patterns. The volume of a 40µm sphere is 1/670th the size of a 350µm sphere. Passing a representative sample of oil through the sensor should give far better information about a failure. Field trials have shown that partial flow sensors can even detect changes in wear patterns due to operating conditions and usage. Weather conditions that generate high levels of wear can be identified and adjustments made to possibly prevent weather-related wear in the future, thus maximizing a turbine’s service life.
The installation of such technologies is best done at the manufacturer. Retro-installation may require modifications to a wind turbine’s lubrication systems. This is not always an easy task when working in a nacelle, so allow for an additional cost. Only a few of the advantages and disadvantages of on-line oil sensors have been covered in this article. To cover as many failure modes as possible, a complete condition monitoring package should ideally consist of vibration-analysis equipment, in a lab, and online oil sensors. WPE
Filed Under: Uncategorized
Mette Enøe says
I find those interesting:
3 Viscosity sensor – Viscosity in cSt, 5 Moisture sensor – Dissolved water in %RH.
We would like to measure viscosity in a pectin extract 75C with small lumps ( citrus peel) ,- Do you think that would be possible?
What is the cost of 3 Viscosity sensor & 5 Moisture sensor – Dissolved water in %RH ?
Best regards Mette Enøe R&D